Data for Dual-Probe Transcranial Full-waveform Inversion: a Brain Phantom Feasibility Study

Published: 19 July 2023| Version 1 | DOI: 10.17632/rbx3ybd5zx.1


This dataset contains ultrasound computed tomography (USCT) data for the article "Dual-Probe Transcranial Full-waveform Inversion: a Brain Phantom Feasibility Study" (Robins et al., 2023). This study investigated the application of full-waveform inversion (FWI) for performing transcranial brain imaging by applying FWI to reconstruct a brain-mimicking phantom. Imaging datasets were acquired using a custom dual-probe acquisition system which consisted of a large water tank (0.7 x 0.6 x 0.6 m^3) and a pair of P4-1 cardiac probes rotated about an imaging target to form a ring array (see Experimental_Imaging_Setup.png). All datasets were acquired using the same imaging sequence which involved a ring array of 16 probe positions (see Dual_Probe_Ring_Array.png). Estimated sets of transducer element positions (elementList.txt) and source signals (signal_filtered.mat, seen plotted in Source_Signal_Filtered.png) have been optimised for running FWI to image the phantom. All transducer positions and source signals were calibrated using a watershot dataset, which was acquired by performing the scan without the phantom being present. Paper DOI: Imaging sequence: o Imaging was performed by a pair of P4-1 cardiac probes (ATL, 96 elements per array) . o The ring array consisted of 3072 element positions (1536 over 16 Probe 01 positions, 1536 over 16 Probe 02 positions). o 384 source elements. o Sampling frequency (Fs) = 22.727 MHz. o [Number of time samples (Nt), No. of Traces] = [3328, 442368]. o Source-receiver pairings for all 442368 traces is given by elementRelas.txt. These imaging experiments include: (1) 01_Experimental_USCT_Brain.mat: o USCT data from imaging the 2.5D PVA brain phantom. (2) 02_Synthetic_USCT_Brain.mat: o USCT data from synthetic imaging the 2.5D numerical brain phantom. o The true numerical brain phantom for this imaging problem is given in TrueVp.mat. (3) 03_Experimental_USCT_WaterShot.mat: o USCT data from imaging just water. Additional files include: elementList.txt: Calibrated transducer element positions, [Element ID, x,y,z - positions] elementRelas.txt: The source-receiver transducer element IDs for all traces [Trace ID, Source ID, Receiver ID] signal_filtered.mat: 384 source wavelets for all source elements (filtered by *) TrueVp.mat: True velocity model of the numerical brain phantom (dx = 0.186 mm, dimensions = [957, 958]) * To filter USCT data 'dataset' to match the source signals, apply the following matlab solution (or equivalent): Fs = 22.727e06; N = 100; F01 = 0.4e06; F02 = 2.0e06; [Hd a] = fir1(N,[F01, F02]/(Fs/2)); dataset_filtered = filtfilt(Hd,a,dataset);


Steps to reproduce

To produce this data, two P4-1 cardiac probes (ATL) were connected to an ultrasound acquisition system that could use both 96-element probes as a single 192-element transducer array (in the related study, a Versonics Vantage 256 was used). These probes were rotated within a large water tank by a pair of rotary motors to form a ring array (as seen in Dual_Probe_Ring_Array.png) to image a brain phantom. An imaging sequence consisting of motor steps and imaging events (each consisting of a single transducer element transmitting and all other elements receiving) was then used to scan the imaging target. Of the 1536 elements of the ring array, a subset of 384 evenly spaced elements were used as sources. Imaging targets were held stationary and in the centre of this ring array throughout the scan. To calibrate the positions and source wavelets used to acquire this ultrasound computed tomography (USCT) data, a scan was performed in the water tank without any imaging target. Optimising these parameters was achieved by minimising the misfit between the observed watershot signal data and a numerical watershot dataset acquired by modelling the transmission events performed when imaging. For all other details for this imaging method, including the fabrication of the imaging phantom, please refer to the related article or contact the corresponding author of the publication.


Imperial College London


Brain Imaging, Ultrasound, Ultrasound Computed Tomography, Imaging Phantom


Engineering and Physical Sciences Research Council